1

IN SEARCH OF DIRECT REALISM Laurence BonJour

University of Washington

It is fairly standard in accounts of the epistemology of perceptual knowledge

to distinguish three main alternative positions: representationalism (also called

representative realism or indirect realism), phenomenalism, and a third view that is

called either naïve realism (usually by its opponents) or direct realism (usually by

those who are more sympathetic to it). I have always found the last of these views

puzzling and elusive. My aim in this paper is to try to figure out what direct realism

amounts to, mainly with an eye to seeing whether it offers a genuine epistemological

alternative to the other two views and to representationalism in particular. My

thesis will be that it does not—that what is right about direct realism turns out to

have little bearing on the main epistemological issue concerning perceptual

knowledge.

Both for reasons of space and to allow a focus on the issues that I regard as

the most important, my discussion will be restricted in the following ways: First, I

will give no consideration at all to phenomenalism. The numerous and to my mind

entirely decisive objections to the phenomenalist view are by now very well known.

Second, I will also give no attention to the standard externalist views of justification,

of which reliabilism is the most familiar. Though it is very easy within a reliabilist

framework to defend something that might perhaps be characterized as “direct

realism,” the point of such a characterization and of the alternatives that it

implicitly suggests is largely lost in that context. Third, although I started by

speaking of perceptual knowledge, I will pay no real attention to the concept of

knowledge as such, focusing instead on the issue of whether and how perception

2

yields a good reason or basis for thinking that a belief about the material world is

true—the issue of justification, as at least one version of internalism understands

it.1 Fourth, I will follow the usual philosophical practice of concentrating mostly on

visual perception, adding, as is also usual, that I am reasonably confident that the

resulting discussion could be generalized without any crucial modification to

perception via the other senses. (Perhaps it is worth adding that I am interested in

direct realism as a serious epistemological alternative that offers an account of how

perceptual beliefs are justified, not merely as the denial that this issue need be

taken seriously.)

Thus the main issue to be considered is whether there is a genuine direct

realist alternative to representationalism, one that is at least prima facie viable,

with regard to the sort of reason or basis we have or might have for thinking that

the beliefs about the physical world that result from perception, especially visual

perception, are true. (Hereafter the terms “justification” and “justified” will be used

exclusively in the indicated internalist sense that pertains to the possession of a

good reason or basis for thinking that a belief is true.)

I I begin with an elaboration and clarification of the contrast between direct

realism and representationalism, starting with the latter. In first approximation,

the representationalist holds two main theses: first, that what is perceived directly

or immediately in sensory experience is not “external” physical or material objects,

but rather things that are mental or subjective in character—sense-data or sensa,

according to the most standard versions of the view; and, second, that the only

available (reasonably cogent) reasons deriving from perception for thinking that

perceptual beliefs about the physical world are true depend on inference from facts

3

about these directly perceived mental or subjective objects, i.e., from facts about the

character and contours of subjective sensory experience, to conclusions about

physical or material objects. (Notoriously, what “direct” and “immediate” mean in

these formulations is far from obvious and will have to be considered further

below.)2

What then does direct (or naïve) realism amount to? The direct realist

should, I suggest, be viewed as rejecting both of the characteristic theses of

representationalism, while still maintaining that perception is indeed a source of

justification for beliefs about the material world. This commits the direct realist to

holding both: (a) that in normal cases of perception, physical or material objects are,

in a sense admittedly still in need of clarification, directly or immediately perceived;

and (b) that the justification for beliefs about such objects that results from

perception does not depend on the sort of inference from the subjective character of

perceptual experience to which the representationalist appeals, but can be accounted

for in some other way. It is obviously the former of these theses that gives direct

realism its label; but the latter thesis is far more important from an epistemological

standpoint and provides the main dialectical motivation for the view. For the main

alleged virtue of direct realism has been precisely that it avoids the need for the

allegedly difficult and perhaps even hopeless inference to which the

representationalist view appeals.

Here is it important to see clearly from the beginning that an adequate case

for the second direct realist thesis cannot be made merely by raising problems for

the representationalist’s opposing view. No matter how difficult or even seemingly

impossible the representationalist’s attempted inference from subjective experience

4

to the material world may turn out to be, this is not enough by itself to show that

direct realism provides a better epistemological alternative. Thus the only way to

adequately defend the second direct realist thesis is to actually give a competing

account of how perceptual beliefs about material objects are justified, presumably by

appealing in some way to the direct or immediate experience of such objects that is

alleged in the first direct realist thesis. And until and unless such an account is

given, a direct realist victory with regard to the issue defined by the opposing first

theses would amount to very little, at least from an epistemological standpoint, since

it would still leave the representationalist’s account of reasons for perceptual beliefs

as the only apparent contender in the field (phenomenalism aside), however

allegedly problematic it may be.

From this perspective, it is a striking fact that many of the recent

proponents of direct realism offer little or nothing by way of a positive account of

how perceptual beliefs are justified according to their view, concentrating instead on

defenses of their first thesis and on arguments against the two representationalist

theses. Thus, for example, George Pitcher, in his defense of direct realism in his

book A Theory of Perception, expresses the assumption that direct realism “is the

most desirable sort of perceptual theory from an epistemological point of view,”3 but

never gives any real explanation of how and why this is supposed to be so, of just

what the direct realist epistemological account amounts to. Similarly, in D. M.

Armstrong’s earlier book Perception and the Physical World,4 arguments are

mounted at great length against representationalism (and against phenomenalism),

focusing mostly on the issue defined by the opposing first theses. But the only hint

of a positive account of how perceptual beliefs are justified or rationally acceptable

5

consists in an appeal to the regress argument for some form of foundationalism,

leading to the conclusion that “There must be at least some truths that we know

without good reasons,” together with the further remark that

. . . since immediate perception is, avowedly, the court of last

appeal when it comes to questions about physical reality, there is

no objection, it seems, to saying that in immediate perception at

least, we acquire knowledge of certain facts about the physical

world without good reasons.5

This may point in the direction of Armstrong’s later externalist view,6 but it offers

no help at all in our search for an internalist account of how perceptual beliefs are

justified according to direct realism, saying at most, perhaps, that such an account

ought from an intuitive standpoint to be available.

Many other recent defenses of direct realism do no better in this regard.

Jonathan Dancy, in his book An Introduction to Contemporary Epistemology,7

ostensibly opts for direct realism for reasons that are again based mainly on

supposed problems with the competing views, and then somewhat surprisingly

suggests a coherentist account of how perceptual beliefs are justified.8 For reasons I

have discussed elsewhere,9 I am extremely doubtful that any such view can succeed.

But for now it is perhaps enough to point out that it really amounts to a denial that

perceptual experience as such offers any justification for the beliefs that result from

it, rather than an account of how such justification works, so that Dancy’s view is

not really a version of direct realism as I am understanding it here. John Searle, in

his book Intentionality,10 argues for the intentionality of visual experiences, claiming

that such experiences both are about material objects and constitute presentations

6

rather than mere representations of such objects, an idea about which I will have

more to say shortly. He also argues briefly against phenomenalism and

representationalism, though his arguments against the latter view are not

particularly compelling. But to the issue of what reason or basis we have on the

basis of perceptual experience for thinking that this intentional, presentational

content is true, Searle has virtually nothing to say. And while both Hilary Putnam

and John McDowell offer views that seem to be intended as versions of direct

realism,11 it is again hard to find in either of their accounts any real response to the

issue of how perceptual beliefs are justified.

II I turn to a consideration of the issue defined by the first of the direct realist

and representationalist theses. Here it is very easy to make a case for the

correctness of the direct realist thesis on at least two fairly obvious understandings

of the terms “direct” or “immediate.”

Consider, first, the question of whether our perceptual awareness of material

objects is arrived at via explicit inference from either beliefs about or awarenesses of

subjective entities such as sense-data. Certainly some representationalists,

especially earlier and less cautious ones, have said things that suggest such a view.

But it is entirely too obvious to be denied that in ordinary cases there is no

consciousness at all of such an inference—and that appeals to “unconscious

inference” are mere evasions of the question. In most ordinary situations, it is

material objects and situations that are the primary or usually the exclusive objects

of the perceiver’s explicit awareness and thought, with no hint that this awareness

has been arrived at via a transition from anything else. In fact this point is so

obvious as to make it hard to see how an explicit inference view could be plausibly

7

ascribed to any serious philosopher, and for this reason, it seems to me preferable

not to view even representationalists as committed to such a claim. This is not to

say that the perceiver is not also conscious in some way of, in the visual case, the

patterns of color and shape that philosophers have characterized in terms of

subjective sense-data or adverbial feature. But this is normally very much in the

background and not at all the object of explicit attention.

Second, as Searle seems to have in mind in speaking of “presentation,” there

is in addition an obvious and intuitively compelling way in which my perceptual

experience seems to directly present physical objects and situations to me: given

suitable conditions of lighting and so forth, I have only to open my eyes, and there

they are, in a way that seems completely open to my cognitive access. I find it

somewhat difficult to say clearly what this direct presentation of physical objects

really amounts to, but it is undeniably a feature of the phenomenology of perceptual

experience, one that would not automatically follow from its non-inferential

directness alone. (It seems reasonably clear that ordinary beliefs about material

objects are less than fully direct, in that they are based on—though not in any clear

way inferred from—these more basic presentational awarenesses.)

We thus have at least two reasonably clear senses in which the perceptual

awareness of material objects can be said to be direct, thereby establishing a case for

the first of the two direct realist theses. But what about the second thesis, which I

have argued to be epistemologically more fundamental? Do either of these ways in

which perceptual awareness of material objects is direct yield any clear alternative

to the representationalist’s account of how claims about such objects must ultimately

be justified?

It is easy to see that the non-inferential character of perceptual awareness

8

says nothing positive about the issue of justification, but merely rules out one way in

which the resulting claims might have been justified. A sheer hunch or random

thought is after all also non-inferentially arrived at, but that plainly constitutes no

reason for thinking that it is correct and so no basis for justification (though we will

consider below a view that seems to dispute this).

What about the presentational directness of perceptual awareness? It is easy

to see how the way in which material objects seem to be simply presented or, as one

might even be tempted to say, given in perceptual experience could lead to the view

of at least the most naïve level of common sense that there is no problem at all about

the justification or indeed about the truth of the resulting beliefs. But as long as the

presentational character of perceptual experience is not confused with the more

Cartesian version of immediacy that I will consider momentarily, it is far from clear

why it should be accorded any genuine justificatory force. This presentational

character has to do with the way in which physical objects are represented or

depicted in experience, but has no obvious bearing on whether such representations

or the beliefs that reflect them are true. A presentational representation is no doubt

more vivid, more striking, in something like the way in which a picture is more

compelling than a merely verbal description. But pictures are just as capable of

being mistaken as anything else,12 and so the pictorial character of a representation

seems to be simply irrelevant to the issue of justification; my suggestion is that we

have so far seen no clear reason not to say the same thing about the presentational

character of perceptual experience.13

In thinking about the foregoing point, it is useful to explicitly contrast the

presentational directness or immediacy just described with the further, more

9

controversial notion of directness or immediacy that has been advocated by

epistemologists in the broadly Cartesian tradition. According to this Cartesian view

of immediacy, sometimes referred to as “acquaintance” or “givenness,” mental and

sensory states and the objects that they are taken to involve are directly accessible

to consciousness, directly before the “eye” of the mind. This view has often been

accompanied by a claim that mistake about them is entirely impossible, but it is

important to see that this is at best a consequence of the main view, not its central

core. The central idea, I suggest, is rather that such entities and features are simply

part of the content of consciousness itself, so that the most fundamental

consciousness of them, that by virtue of which they exist at all, could not indeed be

mistaken. Even if this were so, it would not preclude the formation of mistaken

beliefs about this content as a result of confusion or inattention or bias of some sort,

but would still provide a clear basis for justification, in that the content of such a

belief could be directly compared in consciousness to the conscious feature or object

that it purports to describe.

Notoriously many recent epistemologists, including most of the proponents of

direct realism, have been highly critical of this Cartesian picture, skeptical about

whether anything is genuinely given to consciousness in the way that it claims. My

own view, developed elsewhere, is that the Cartesian view is defensible and even

obvious if properly understood.14 But my point for the moment is that whether or

not such a view is defensible in relation to our awareness of the features or aspects

of sensory and mental states, it is plainly not applicable to our perceptual awareness

of material objects. Material objects, understood in a realist, non-phenomenalist

way, are plainly outside the mind, metaphysically distinct from any sort of

10

experience or awareness of them, and related to conscious experience only via a

highly complicated causal chain. They are thus inherently incapable of being

directly given to consciousness in the way that things like sense-data are claimed by

the Cartesian to be—a point that should, of course, be readily accepted by those who

think that the Cartesian brand of directness or immediacy is simply a myth and so

does not genuinely apply to anything. This means that even the most fundamental

perceptual awarenesses of material things are entirely distinct from their objects in

a way that allows room for the issues of correctness and of justification to be

significantly raised and precludes the simple response that a Cartesian version of

immediacy might seem to make available.

If the Cartesian version of directness or immediacy should turn out, as I

believe, to be more than a myth, then there would be a clear sense in which

perceptual awareness of material objects is not direct, in addition to the others,

already noted, in which it is direct—so that the issue between the first of the direct

realist and representationalist theses would end in a split decision. But the more

important point is that even if the senses in which the perceptual awareness of

material objects is direct and immediate are the only viable ones, we have so far

found no way in which this result provides any clear support for the crucial second

thesis of direct realism. Even if the perceptual awareness of material objects is

arrived at without inference, and even if the objects in question are from a

phenomenological standpoint simply presented in our perceptual experience, neither

of these forms of directness or immediacy seems in any very obvious way to yield a

good reason or basis for thinking that the resulting claims about the physical world

are true or likely to be true. It thus remains so far quite possible that the account of

such reasons offered by the second representationalist thesis is the only one

11

available.15

It will not have escaped notice that I have arrived at this result with very

little consideration along the way of specific direct realist views. I have already

noted that many such views say almost nothing about the crucial issue of

justification, but there are others that do address that issue more explicitly, and I

now want to have a look at some of these. A full consideration any of them would

require much more space than can be allotted here, but fortunately their responses

to the key issue of justification can be dealt with fairly briefly.

III Perhaps the simplest and most straightforward response to the issue of how

perceptual beliefs about material objects are justified is offered by Michael

Huemer.16 According to Huemer, perceptual experiences are “assertive mental

representations,” where what is thus represented is some sort of specific physical

object or situation. When there is genuine perception, such experiences count as

awarenesses in Huemer’s defined sense, in that (i) there exists an object or situation

that satisfies the content of the experience, and (ii) it is non-accidental (as a result of

a non-deviant causal chain) that this is so [79-80]. And such awarenesses are direct,

he claims, because they are not based upon any other sort of apprehension [80],

where one apprehension A is based on another apprehension B just in case B causes

A because B appears to logically support A [56]. Perceptual beliefs, involving

conceptual claims about physical objects and situations, are based upon such

perceptual experiences, but not inferentially, since inference is a relation between

beliefs, and perceptual experiences are not beliefs: though they have propositional

content (that is, are true or false), that content is not conceptually represented

[74].17 Though there are several questions that might be raised about this view, it

12

seems to me reasonably plausible so far, for reasons that we have already seen:

perceptual experience does seem to involve presentations of material objects that are

capable of being true or false; and these do not seem, at least in general, to be

arrived at on the basis of other apprehensions.18

But all this only raises, or at least appears to raise, the further issue of

whether and how the perceptual experience, or perhaps rather the assertive

representation that it involves, is itself justified: that is, of what reason if any there

is to think that this representation is true or correct. To this pivotal question,

Huemer offers, as I see it, two rather different responses, one directly and one by

implication, neither of them in my judgment adequate.

The first, more direct response is to deny outright that the issue of

justification intelligibly arises for perceptual experiences. According to Huemer, “It

does not make sense—it is a category error—to say that an experience is justified or

unjustified” [97]. People simply have experiences, but cannot be either justified or

unjustified in having them because they are simply “automatic responses to external

stimuli” [97]. But this is almost entirely misdirection. While it is of course true that

people cannot help having perceptual experiences, so that it cannot be intelligibly

said that they should not have had them or are somehow blameworthy for having

them, none of this has any real relation to the main epistemological issue with which

we are concerned. On Huemer’s own account, such experiences have an assertive

representational and indeed propositional content: they represent various specific

objects and situations as existing in the world. But then we can surely ask quite

intelligibly what reason or basis, if any, there is for thinking that the

representational content in question is true or correct, that is, for thinking that the

13

world actually contains the objects or situations thus represented—and this just is

the question of whether the perceptual experience, or more specifically its

representational content, is justified in the sense I have indicated.

Huemer, to his credit, recognizes that this first response to the question of

the justification of perceptual experiences will not do by itself. He thus advances a

further account of why beliefs that are based on perceptual experiences are

themselves justified, even though the experiences themselves are neither justified

nor unjustified. This account appeals to a principle called “the rule of Phenomenal

Conservatism”

If it seems to S as if P, then S thereby has at least prima facie

[that is, defeasible] justification for believing that P. [99]

Here the relevant seeming is the perceptual experience itself.19

But is this principle, which Huemer claims to be a self-evident necessary

truth, in fact correct? I do not have space here to consider everything that Huemer

says in support of it, but the main lines of his defense can, it seems to me, be

captured in the following two claims: First, the principle represents “the

epistemological default position,” in that seemings or appearances are “presumed

true, until proven false,” which he compares to the legal situation in which a

criminal defendant is “presumed innocent, until proven guilty” [100]. Second, “all

thought and reasoning presupposes the principle in a certain sense” [105], in that

there is ultimately no alternative but to rely on what seems to be true, no other

place for reasoning and dialectical criticism to start. And the problem is that neither

of these claims speaks nearly clearly enough to the fundamental issue of whether

there is any reason or basis for thinking that the seemings in question are likely to

14

be true.

This is a bit easier to see with respect to the first claim, where it seems

obvious that neither the computer metaphor (for that is really all it is) nor the really

quite irrelevant legal analogy provides any real insight as to why the claims in

question should be “presumed true” if this is supposed to mean that there is a good

albeit prima facie reason for thinking that they actually are true. Though others

have also made similar suggestions, it just isn’t clear why there should be an

“epistemological default position” of this sort or why there should be an allocation of

the burden of proof that favors the truth as against the falsehood of such a claim.

(And if the presumption is supposed to be reasonable in some other, non-epistemic

way, then it seems simply irrelevant to the issue of epistemic justification.)

As regards the second claim, I am inclined to agree with Huemer that in the

end we have no alternative in our thought and reasoning but to rely on things that

seem to us to be true. Any sort of reasoning or discursive justification must start

somewhere, and it is far from obvious what other sort of starting point there could

be. The important question, however, is whether all forms of seeming are on a par

in this respect, at least to the extent of constituting prima facie reasons for thinking

that the claims are true, as Huemer’s principle claims, or whether it might instead

be the case that some specific ways of seeming constitute reasons of this sort and

others do not. It is easier to argue for the negative part of this latter alternative, for

it surely seems clear, contrary to Huemer, that, for example, a mere spontaneous

hunch provides no reason for thinking that its content is true, no matter how

psychologically compelling it may be. Making a case for the positive part, viz., the

claim that certain specific sorts of seemings do yield genuine reasons, is harder and

cannot be attempted here; I have argued elsewhere that this is true for at least two

15

sorts of seemings, those that involve apprehensions of the contents of one’s own

conscious states of mind and those that involve apprehensions of self-evident

necessary truths.20

But my main point for the moment is that until and unless such a case can be

made, either for some particular sorts of seemings or for all, the fact that we

ultimately have nothing else to rely on does not show that the claims that we accept

on this basis are likely to be true or that we have any good reason to think that they

are. That we have nothing better to rely on, even if this is (as I doubt) genuinely the

case, does not show that the best we have is good enough. Thus if Huemer’s defense

of the reliance on perceptual experience is the best that we can do, the result, I

submit, is skepticism about the resulting claims rather than genuine justification.21

(It is also worth noting the rather surprising fact that the distinctively perceptual

character of the experience really plays no role in Huemer’s account of how the

resulting claims are justified.22)

IV A quite different account of the justification of claims about material objects

within a direct realist view is offered by Bill Brewer.23 Brewer claims to be offering

an account of how perceptual beliefs are justified in an internalist sense, while

avoiding the traditional foundationalist (and representationalist) appeal to inference

from introspective knowledge of the character of subjective experience, an appeal

that he regards as hopeless.24

Brewer’s view can be viewed as an attempt to base the justification of such

claims on the distinctively presentational character of perceptual experience. The

initial idea is that perceptual experience is essentially perspectival and egocentric in

character: it presents physical objects and situations as standing in a set of spatial

16

relations that are centered on the perceiver [43]. Moreover, Brewer argues, these

presented spatial relations are in fact essential to picking out specific physical

objects in a way that makes determinate beliefs about those particular objects even

possible, since merely qualitative descriptions of objects, however complicated they

might be, could never guarantee unique reference [44-45]. Instead such specific

reference must be fundamentally demonstrative in character, appealing to the

egocentric spatial relations presented in experience.

So far, this seems to me plausible enough, but what does it have to do with

the issue of justification? Brewer’s claim is that

. . . egocentric spatial perception displays its objects as

epistemically accessible to the perceiver. . . . in veridical perception,

we are presented with things “out there” in such a way that we

grasp immediately how we are right about the way they are. [43]

Instead of the classical foundationalist’s reliance on inference,

The internalist requirement is met directly instead, by the

presentation of objects of experience as accessible to us, in virtue of

the egocentric spatial relations between them and us which are

displayed in experience. [43-4]

Here it should be noted that Brewer’s somewhat idiosyncratic version of internalism

in application to perception is that “the perceiver must know how he is right about

the way things are in the world around him” [41].25

How are we to understand this? It is reasonable enough to say that

perceptual experience seems to present objects as accessible to us, even though the

full story of how they are thus accessible (if indeed they are) and of how we are right

17

(if indeed we are) seems to demand further elements of a broadly causal sort that

are not in any clear way reflected in the presentational experience itself. But the

main problem, as the “seems” and the parenthetical qualifications (which are mine,

not Brewer’s) suggests, is that nothing about these facts of phenomenology offers

any apparent reason that counts against the possibility that the experiences in

question are deceptive or illusory in a way that would mean that the objects that

they seem to present do not genuinely exist. If this possibility were realized, and if

the rest of Brewer’s view is correct, then the upshot would be that the beliefs whose

attempted reference to objects is based on those experiences in fact fail to refer to

specific objects at all. And if this were the case for all of our perceptual experiences,

then the result would be that we fail to have any determinate beliefs about

particular objects in the world. This is no doubt an intuitively implausible result,

but it cannot be assumed to be false in the present context26 without begging the

essential question. Thus Brewer does not seem to have given any account of why

perceptual experience as such yields any justification or reason of an internalist sort

for thinking that the material objects that it seems to present are genuinely there.

The key assumption underlying the foregoing objection is that veridical

perceptual experiences, those for which the objects that seem to be presented are

really there, and non-veridical experiences, those for which they are not, might be

entirely indistinguishable from each other from a subjective standpoint,

indiscernible to the people who have them, in which case the occurrence of such an

experience would in itself yield no internalist reason or basis for thinking that the

actual situation is one way rather than another.27 Brewer, however, rejects this

assumption. He grants that the two sorts of experience cannot be distinguished in

18

an infallible or indubitable way, but argues that justification does not demand

infallibility or indubitability. And if such a demand is set aside, then, he argues, the

person whose experience is veridical can be justified in the way already described

and hence can know that the objects in question are genuine and hence also know on

that basis that his experience is veridical; whereas such justification and knowledge

is not, of course, available to the person whose experience is non-veridical (even

though it will mistakenly seem to him that it is). As Brewer says in a reply to

Richard Fumerton:

When having a vivid hallucinatory experience, a person does not

have the same reasons for empirical belief as she does when she is

actually perceiving the way things are in the world around her.

For, in the veridical case, her epistemic openness to the way things

are out there is evident to her, whereas, in the hallucinatory case,

she merely seems to be open to the way the world is.28

As he suggests at the end of this reply, his view involves “a more radical type of

externalism” than epistemic externalism of the reliabilist sort (namely, I take it, an

externalist view of content), one in which “a person’s subjective-experiential

condition in perception is itself constituted by her relations with the mind-

independent things in the world around her.”29

It is easy, however, to see that this move will not do, at least if it is a

genuinely internalist reason for thinking that material objects exist that is in

question. The problem, of course, is that even if it is granted that Brewer has

specified a difference between a veridical perceptual state and a non-veridical one,

this difference is not discernible to the individuals in question in the way that an

19

internalist justification requires. Contrary to Brewer’s suggestion, such a claim

need not be based on a demand of infallibility or indubitability. The problem is

rather that there need be no internally detectable difference of any sort for the

persons in question to appeal to, not just one that is infallible. True, the person who

has the veridical experience can claim correctly to be genuinely open to the world

and so to have the sort of reason for and consequent knowledge of the veridicality of

his or her experience that Brewer specifies; while the person who has the non-

veridical experience will in fact be incorrect in making such claims. But it remains

the case that if the issue were explicitly raised, neither of the persons would be able

to tell in a subjectively accessible way on the basis of the presentational character of

the experience alone which situation he or she is in. Thus each of them not only

fails to have what Brewer calls a “skeptic-proof guarantee of truth” [52], but in fact

fails to have any genuinely internalist reason at all. Again we seem to be left with a

skeptical result that is not mitigated to any serious extent by the fact that we may,

for all we can tell, be able to make claims of justification and knowledge that are

correct in senses that we cannot genuinely recognize.

V Both Huemer and Brewer attempt to rest the justification of claims about the

material world on what might be described as full-blown perceptual experience: the

perceptual experience that seems to present physical objects and situations in

physical space. Huemer regards such experience as propositional even though non-

conceptual, whereas Brewer regards it as conceptual and presumably also as

propositional. A third, quite different view, but still a version of direct realism as

understood here, is rather tentatively defended by Steven Reynolds.30 Reynolds

attempts to base the justification of perceptual beliefs on perceptual experience

20

conceived in a thinner, more austere way: roughly, in the visual case, as the

experience of patches of color of various sizes, shapes, and hues, arranged in a

person’s “visual field.” He characterizes such experiences as non-propositional and

non-intentional, and I believe that he would say that they are non-conceptual as

well.

This is essentially the same sort of experience that the most standard forms

of representationalism attempt to characterize in terms of sense-data and the like

and appeal to as the basis for their envisaged inference to the “external world.” I

have already said that I think that a consciousness of this sort of experience

normally exists in ordinary perception, albeit very much in the background, and it is

interesting to speculate as to how it relates to the richer level of experience to which

Huemer and Brewer appeal. My tentative suggestion would be that the richer level

of experience is in some way an amalgam of this thinner sort of experience with

conceptual elements: what C. I. Lewis, among others, used to describe as the

“conceptual interpretation of the given.” (Though how these elements are fused into

a phenomenologically seamless result is not easy to say.) Reynolds attempts to

explain how beliefs about material objects can be justified by non-propositional

experiences of this sort, but I think that the account, if successful, would apply just

as well to the justification of the richer variety or level of perceptual experience.

Reynolds suggests that the claims about material objects that we arrive at on

the basis of non-propositional experience reflect learned recognitional skills, skills

that are at least somewhat analogous to those involved in such things as playing the

piano or speaking a natural language [282]. Like other skills, these recognitional

skills could be captured in rules, though the person who has the skills need not be

thought of as having any very explicit conception of these rules. I think that this

21

picture is at least largely correct, that in arriving at claims about the material world

in response to the non-propositional level of experience we are guided by a tacitly

understood correlation of some complicated sort between experiential content and

material situations, with the only real question being whether this correlation is, as

Reynolds suggests, learned, or whether it is not instead partly or even largely

innate.

But what are the implications of this picture for the justification of the

resulting beliefs (and propositional perceptual experiences)? Reynolds’s suggestion

is that the resulting claims are epistemically justified just in case they in fact

conform to the rules that describe the recognitional skill in question, that is just in

case they fit the correlation between non-propositional experience and material-

object claims that (we confidently think) is shared within our epistemic community.

In his view, the epistemic goal of truth comes in only as part of the causal

explanation of why we adopt the practice of belief acquisition that involve these

rules: it is adopted because following such rules in fact tends to produce true beliefs

[288].

But this account is not enough to yield a genuinely internalist alternative to

representationalism (something, to be sure, that Reynolds makes no very explicit

claim to be doing). I have no real quarrel with saying that there is a species of

justification that results from conformity to this tacitly accepted correlation or set of

rules, something perhaps close to what Tyler Burge and others have in mind by

speaking of entitlement.31 But whether the justification (or entitlement) that results

in this way is epistemic in character must surely depend in the end on whether or

not the correlation between experience and material objects or situations reflected in

22

the correlation in question is indeed accurate or reliable in the sense that the claims

about the material world arrived at on that basis are in fact likely to be true. And

for the resulting justification to be genuinely internalist in character, it seems that

there must be good internalist reasons for thinking that this is so. Thus, I suggest,

we are led inexorably back to the classical representationalist project of trying to

show how an inference from the features of non-conceptual and non-propositional

experience to claims about the material world can be justified. Perhaps ordinary

people need not actually be able to make this inference in order for them to be in

some sense entitled to their beliefs, but that it be possible to make it is, I am

suggesting, a necessary condition for this entitlement to count as genuinely

epistemic, from an internalist standpoint.

VI I have now examined, in the previous three sections, three specific accounts

of how perceptual claims about the material world might be justified within a direct

realist view in ways that differ from that advocated by the representationalist and

have argued that none of these accounts is successful. These results obviously do

not in themselves establish that there is not some other possible direct realist

account that is more tenable, even though I have been unable to find one. But I

want to suggest briefly that there is some reason, though hardly a conclusive one, for

thinking that any further such account will be in the vicinity of one of the ones just

considered and likely open to the same objections.

The reason in question is based on the idea that versions of direct realism can

be usefully distinguished by what they say about the nature of perceptual experience

and its role in justification. If such experience is to play a justificatory role, then the

direct realist must seemingly either (i) ascribe to experience a propositional content

23

of an internally accessible sort and appeal to that content for the justification of

perceptual beliefs, or (ii) ascribe to perceptual experience a propositional content

that is partly external in the content-externalist sense and appeal in part to that

external content for the justification of perceptual beliefs, or (iii) attempt to justify

perceptual claims by appeal to only non-propositional experiential content.

Huemer’s account is of sort (i) and faces the problem, in my view insoluble, of how

the subjectively accessible propositional content of the perceptual experience is itself

to be justified. Brewer’s account is of sort (ii) and faces the problem, which I again

regard as insoluble, that the sort of justification it offers will not be accessible to the

believer in the way that a genuinely internalist view requires. And a version of

alternative (iii), such as that advanced by Reynolds, seems to lead directly back to

representationalism in the way suggested. I can see no further possibility that is

compatible with the idea that perception experience is itself somehow a source of

justification. As already suggested, this is not intended to be anything like a

demonstrative argument. But it may suggest that in investigating the specific views

I have considered, more progress has been made toward assessing the overall

viability of direct realism than might otherwise have been thought.

Is there any further direct realist alternative, one that avoids any claim that

perceptual experience has a content that plays a role in justification, while still

claiming that perception is somehow a source of justification? The obvious problem

is that it is extremely unclear where such justification is supposed to come from.

Moreover, I would suggest, it is after all pretty obvious on reflection that

perceptual experience does contribute in an important way to the justification that

perceptual beliefs seem intuitively to have. For example, as I look out on this room

and form various perceptual beliefs concerning the presence of people and chairs and

24

walls and windows, my reasons or apparent reasons for thinking those various

things are true surely depend in a fundamental way on the presence in my visual

field of what might be characterized as distinctively shaped patches of color. These

phenomenal elements seem to present to me the objects in question and even

perhaps to be in some way identical with them. I have argued that this seeming

presentation cannot simply be accepted at face value. But it still is, I suggest,

abundantly obvious that the presence in my experience of elements of this kind

plays an essential role in the only sort of reason or basis for thinking that my

perceptual claims are true that I even seem to myself to have. And once direct

realist views of the sorts already considered in the previous three sections are set

aside, only the representationalist account of how such a reason could be genuinely

cogent seems to remain. The difficulties that face such a view are extremely

familiar and obvious, and I do not mean to take them lightly. But I would also

suggest that far, far too much of the discussion of these issues, from the time of

Descartes and Locke to our own, has been concerned with ways of attempting to

evade the need for such an account and far too little with actually attempting to give

one—with the various views that fall under the rubric of direct realism being only

the latest attempts at such evasion and likely, I think, to be no more successful in

the end than the idealist and phenomenalist views that preceded them.

25

NOTES

1 I am not assuming here that a reason or basis for thinking that a belief is true must

take the form of another belief; on the contrary, anything to which one has cognitive access that provides a rational basis for thinking that a belief is true will count as such a reason.

2 The representationalist does not, of course, deny that material objects are genuinely perceived, but only attempts to give an account of how such perception works and of why it is not direct or immediate in the way that it might seem commonsensically to be. It is also worth noting that while the most standard formulations of representationalism appeal to sense-data (or to entities of arguably the same basic sort, such as the “ideas” of the British empiricists), neither the sense-datum theory nor the more general “act-object” picture that underlies it is in any way essential to the basic representationalist view. If, for example, one adopts instead an “adverbial” account of perceptual experience, the representationalist view can still be formulated as the theses (a) that what one is most directly aware of in perceptual experience is the adverbial features of experiential states (which are, of course, quite distinct from material objects) and (b) that any good reasons we may have for believing that material objects of various sorts exist depend on inference from facts about these adverbial features. (It will be convenient to continue to use the term “object” to refer to whatever it is of which one is directly aware in perception, irrespective of which metaphysical category it may belong to.)

3 George Pitcher, A Theory of Perception (Princeton: Princeton University Press, 1971), p. 217.

4 D. M. Armstrong, Perception and the Physical World (London: Routledge & Kegan Paul, 1961).

5 Ibid., p. 120. As I understand the idea of a reason, Armstrong’s characterization of foundationalism is inaccurate. What the regress argument shows is that there must be beliefs that are justified in some way that does not rely on further beliefs, not that there are beliefs that are justified in the absence of any sort of good reason or basis for thinking that they are true.

6 In his book Belief, Truth and Knowledge (Cambridge: Cambridge University Press, 1973).

7 Jonathan Dancy, An Introduction to Contemporary Epistemology (London: Oxford University Press, 1985). The main discussion of perception is in chapters 10 and 11.

8 Ibid., p. 176 ff. 9 See my paper “The Dialectic of Foundationalism and Coherentism,” in the Blackwell

Guide to Epistemology, ed. John Greco and Ernest Sosa (Oxford: Blackwell, 1999), pp. 117-42.

10 (Cambridge: Cambridge University Press, 1983), ch. 2. 11 See Hilary Putnam, “Sense, Nonsense, and the Senses” (his Dewey Lectures), Journal

of Philosophy, vol. 91 (1994), pp. 445-517; and John McDowell, Mind and World (Cambridge, MA: Harvard University Press, 1994).

12 Putting aside irrelevant causal facts about photographic images in particular. 13 For an opposing view, see James Pryor, “The Skeptic and the Dogmatist,” Noûs, vol.

34 (2000), pp. 517-49. Pryor’s claim, defended on intuitive grounds, is that merely the occurrence of “an experience as of p’s being the case,” where p is one of a large class of

26

propositions about material objects that are “basically represented by our experience,” yields prima facie justification for believing p. But while I agree that there is a commonsense intuition of this sort, I do not think it is enough to obviate the need to explain just how the justification is supposed to work, how the occurrence of such an experience yields a reason or basis for thinking that the proposition in question is true. And in the end, I think that the representationalist’s account, based on the qualitative character of the underlying experience, is the only one that is available. (See the discussion, below, of Reynolds’s appeal to recognitional skill.)

14 See the paper cited in note 9. 15 Though this might be a pyrrhic victory for the representationalist if the awareness of

sense-data (or adverbial features) cannot itself be justified by appeal to something like the Cartesian notion of immediacy.

16 In his book Skepticism and the Veil of Perception (Lanham, MD: Rowman & Littlefield, 2001). Bracketed references in this section are to the pages of this book.

17 Though the relation between a belief and the corresponding experience is not inferential, according to Huemer, it is logical in a broader sense: the non-conceptually represented content of the perceptual experience is either the same as or, normally, a more determinate version of the conceptually represented content of the belief, so that the former entails the latter.

18 Many are likely to have doubts about the claim that perceptual experiences have propositional content without being conceptual; indeed, I have heard it claimed that such a view is simply incoherent, that only representations that involve concepts can have propositional content. In the end, I am inclined to agree that this is so and hence that Huemer’s view on this point is mistaken. But I do not think that this is immediately obvious or that there is any very simple way to argue for it.

19 One thing worth noticing here is that if this principle were correct, and assuming that the justification it speaks of is supposed to be epistemic justification in the sense of there being a good reason or basis for thinking that the claim in question is likely to be true, then it would follow from the principle that a suitably abstract version of the representational content of the perceptual experience would, after all, also be justified in the sense just indicated—which is why I regard this principle as a second, quite different response to the question of how perceptual experiences are justified. (“Suitably abstract” because this point pertains only to the part of the experience’s content that is at the same level of abstraction as the belief and would not extend to the more specific content of the experience, the ways in which it is more determinate than the belief. But it is hard to see how the more abstract content could be justified if this specific content were not.)

20 See the paper cited in note 9; and my book In Defense of Pure Reason (Cambridge: Cambridge University Press, 1998), chapters 4 and 5.

21 To be fair to Huemer, I should note that he seems to be operating with what has become known as a deontological conception of epistemic justification, one according to which a belief is justified if the believer is epistemically blameless in holding it. If there were no way to discriminate among seemings with respect to whether they yield genuine reasons for thinking that the corresponding claims are true, then it would be plausible enough that a person who accepted beliefs on the basis of seemings generally would be epistemically blameless simply because that is the only epistemic alternative that would be open. As I have argued elsewhere, however, being epistemically blameless (or satisfying other similar deontological requirements) is not enough for genuine epistemic justification: the aim of epistemic justification is truth, and the connection between blamelessness and truth is far

27

too tenuous—as indeed the point just made illustrates. (See my paper “The Indispensability of Internalism,” forthcoming from Philosophical Topics.) I have taken Huemer to be primarily concerned with genuine epistemic justification and have interpreted his view accordingly; to take him to be concerned only with epistemic blamelessness would mean that his view fails to be a version of direct realism in the sense explained above, i.e., no longer a genuine epistemological alternative to representationalism.

22 Making the “Phenomenal” part of the label for the principle somewhat misleading. 23 In his book Perception and Reason (Oxford: Oxford University Press, 1999) and in

his earlier article “Foundations of Perceptual Knowledge,” American Philosophical Quarterly, vol. 34 (1997), pp. 41-55. For reasons of space, I will mostly follow the simpler version of the view presented in the article (bracketed references in this section are to its pages), thus leaving aside many further interesting details, arguments, and ramifications, which do not, however, in my judgment affect the central issue I am concerned with. See also a book symposium on Perception and Reason in Philosophy and Phenomenological Research, vol. 63 (2001), pp. 405-64.

24 See the discussion in Perception and Reason, pp. 112-29. 25 In Perception and Reason, Brewer offers a further argument (in chapter 3) for the

thesis that the same relations between perceptual experiences and perceptual beliefs that determine the specific content of those beliefs must also yield reasons for thinking that those beliefs are true. This argument seems to me objectionably verificationist in character, but I have no room to discuss it here.

26 As Brewer sometimes seems to do. See, e.g., Perception and Reason, pp. 20-21. 27 I do not mean to suggest here that there might not be further reasons of some sort for

favoring one of these alternatives over the other; but such further reasons would have to appeal to more than the experiences themselves.

28 Brewer, “Reply to Fumerton,” Philosophy and Phenomenological Research, vol. 63 (2001), p. 451.

29 Ibid., p. 452. My italics. 30 Steven L. Reynolds, “Knowing How to Believe with Justification,” Philosophical

Studies, vol. 64 (1991), pp. 273-92. Bracketed references in this section are to the pages of this article.

31 See Tyler Burge, “Content Preservation,” Philosophical Review, vol. 102 (1993), pp. 457-488. Burge cites Reynolds’s view as at least in the vicinity of his own (p. 478), but suggests that he would appeal to “the perceptual character of perceptual belief” rather than to non-intentional sensations. This may lead back to something like Brewer’s view, but Burge does not say enough to make it possible to be very sure.